MPC-ABCO: An MPC-Based Adaptive Bezier Curve Optimization Framework for UAV-UGV Cooperative Landing

Abstract

This letter addresses the approach of enabling a UAV to autonomously land on an agile UGV that moves irregularly with sudden accelerations, decelerations or abrupt turns, where existing methods have limitations in handling nonlinear dynamics and visual recognition errors in high-dynamic situations. To overcome these limitations, we present a real-time UAV-UGV cooperative landing framework that integrates Model Predictive Control and Adaptive Bezier Curve Optimization (MPC-ABCO) to achieve UGV's position prediction and UAV's trajectory forecasting, ensuring robust performance in high-dynamic scenarios. MPC-ABCO utilizes an MPC model to predict the UGV's future states by leveraging odometry data to model UGV's kinematic dynamics. Concurrently, the UAV's landing trajectory is optimized by Bezier curves, with waypoints dynamically updated by considering UGV's relative future positions detected by AprilTag code. Experimental results demonstrate that the proposed framework achieves an average landing deviation below 5 cm for UGV's steering velocities up to 5 m/s under no-acceleration stations. When the UGV follows irregular paths with random accelerations, MPC-ABCO outperforms traditional strategies in both landing success rate and accuracy, maintaining reliable performance at maximum velocities of 5 m/s.